Regulation of Cross-linking of Actin Filament by IQGAP1, a Target for Cdc42*

We have previously shown that IQGAP1, a recently identified target for Cdc42 and Rac1 small GTPases, showed a distribution similar to that of cortical actin cytoskeleton at the membrane ruffling area induced by insulin and Rac1val12 (Kuroda, S., Fukata, M., Kobayashi, K., Nakafuku, M., Nomura, N., Iwamatsu, A., and Kaibuchi, K. (1996) J. Biol. Chem. 271, 23363–23367). Here we identified an IQGAP1-interacting molecule with molecular mass of 43 kDa (p43) from bovine brain cytosol, using glutathione S-transferase (GST)-IQGAP1 affinity column chromatography. The amino acid sequencing of the protein revealed that p43 was identical to β- and γ-actin. IQGAP1 was cosedimentated with filamentous actin (F-actin). The amino-terminal domain (amino acids 1–216) of IQGAP1 was responsible for the interaction with F-actin. Falling ball viscometry assay revealed that IQGAP1 cross-linked the F-actin. This IQGAP1 activity was further enhanced by guanosine 5′-(3-O-thio)triphosphate (GTPγS)·GST-Cdc42 but not by GDP·GST-Cdc42. The gel filtration analysis of IQGAP1 revealed that IQGAP1 appeared as oligomers and that GTPγS·GST-Cdc42 but not GDP·GST-Cdc42 enhanced the oligomerization of IQGAP1. These results strongly suggest that IQGAP1, acting downstream of Cdc42, can cross-link the actin filament through its oligomerization.

shown to be involved in the regulation of actin reorganization (10,16). Despite of these intensive studies, how Cdc42 and Rac1 regulate actin cytoskeleton is still unclear. We have previously identified IQGAP1 as a target for Cdc42 and Rac1 and found that it was colocalized with the actin filament at the lamellipodia induced by insulin and Rac1 val12 (14). This result led us to examine the role of IQGAP1 as a regulator of actin cytoskeleton.
In this study, we identified actin as an IQGAP1-interacting molecule and found that IQGAP1 interacted with F-actin in vitro and cross-linked F-actin in a GTP␥S⅐GST-Cdc42-dependent manner. We also showed here that IQGAP1 formed oligomers, suggesting that IQGAP1 cross-links the actin filament through its oligomerization.

EXPERIMENTAL PROCEDURES
Materials and Chemicals-Anti-actin polyclonal antibody was kindly provided by Dr. I. Yahara (Tokyo Metropolitan Institute of Medical Science, Tokyo, Japan). Anti-IQGAP1 polyclonal antibody was raised against GST-IQGAP1 (aa 1-863) as an antigen. All materials used in the nucleic acid study were purchased from Takara Shuzo Co. (Kyoto, Japan). Other materials and chemicals were obtained from commercial sources.
Preparation of Recombinant Protein-Myc-IQGAP1 (aa 1-1657) and GST-IQGAP1 (aa 1-1657) were purified from overexpressing Spodoptera frugiperda insect cells. The insect cells overexpressing Myc-IQGAP1 were homogenized with Buffer A (20 mM Tris/HCl at pH 7.4, 1 mM dithiothreitol, 1 mM EDTA, and 10 M (p-amidinophenyl)-methanesulfonyl fluoride on ice and centrifuged at 100,000 ϫ g for 1 h at 4°C. For cosedimentation assay, the pellets were extracted by the addition of Buffer A containing 500 mM NaCl. After shaking for 1 h at 4°C, the extracts were centrifuged at 100,000 ϫ g for 1 h at 4°C. Then Myc-IQGAP1 was purified from the supernatant by MonoQ column chromatography. For falling ball viscometry assay, GST-IQGAP1 was purified by glutathione-Sepharose column chromatography from overexpressing * This study was supported by grants-in-aid for Scientific Research and for Cancer Research from the Ministry of Education, Science, and Culture of Japan (1996) and by grants from Kirin Brewery Co. Ltd. and from the Mitsubishi Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Gel Filtration Chromatography-Myc-IQGAP1 (25 g) was incubated at 4°C for 1 h with or without GDP⅐GST-Cdc42 or GTP␥S⅐GST-Cdc42 (100 g each) in Buffer E (20 mM Tris/HCl at pH 7.4, 1 mM dithiothreitol, 6 mM MgCl 2 , 300 mM NaCl, and 1.9 mM EDTA). Superose 6 HR 10/30 (Pharmacia Biotechnology Inc.) gel filtration column was equilibrated with the Buffer A containing 200 mM NaCl. Each sample was centrifuged at 10,000 ϫ g for 1 h at 4°C. The supernatant was loaded onto the column at a flow rate 0.25 ml/min, and 0.25-ml fractions were collected. An aliquot of each fraction was subjected to SDS-PAGE followed by Western blotting (30).
Other Procedures-The amino acid sequencing of p43 protein was carried out as described (31).

RESULTS
We and other groups have recently identified IQGAP as a target for Cdc42 and Rac1 (11)(12)(13)(14). We found that IQGAP1 accumulated at the membrane ruffling area induced by Rac1 val12 and by insulin in KB cells, where cortical actin cytoskeleton was localized (14). To understand the function of IQGAP1, we attempted to identify an IQGAP1-interacting molecule(s). Bovine brain cytosol was subjected to the GST, GST-IQGAP1-N-1 (aa 1-863), or GST-IQGAP1-C-1 (aa 764 -1657) coated affinity column. After washing the columns, the proteins bound to GST-IQGAP1 were eluted with 500 mM NaCl in Buffer A. Proteins with molecular masses of 43 (p43) and 41 kDa (p41) were specifically detected in the GST-IQGAP1-N-1 (aa 1-863) eluate (Fig. 1A). These proteins were not observed in either the eluate from GST or GST-IQGAP1-C-1 (aa 764 -1657). Therefore, these proteins are likely to be IQGAP1-interacting molecules. To determine the molecular identity of p43, it was subjected to the amino acid sequencing. The peptide sequences derived from p43 were EITALAPSTMK and AGFAG-DDAPRAVFP, both of which are identical to those of ␤and ␥-actin. The molecular weights of ␤and ␥-actin were calculated to be 41,605 and 41,661, respectively, which are almost the same as that of p43. We also confirmed that p43 was recognized by anti-actin antibody (Fig. 1B). Therefore, we concluded that p43 was actin. Identification of p41 is currently under investigation. 2 IQGAP1 has a CHD (17), which is also present in several actin-binding proteins, including calponin, filamin, ␣-actinin, and fimbrin, in its amino terminus (13,32). To determine whether IQGAP1 directly interacts with F-actin, cosedimentation assay of recombinant IQGAP1 with F-actin was performed. Myc-IQGAP1 was cosedimentated in the presence of F-actin but not in the absence of F-actin ( Fig. 2A), indicating that IQGAP1 directly interacts with F-actin. GST-IQGAP1-N-1 (aa 1-863) was also cosedimentated with F-actin, whereas IQ-GAP1-C-1 (aa 764 -1657) was not (data not shown). We produced the indicated mutants of IQGAP1, and their interactions with F-actin were assessed. GST-IQGAP1-N-2 (aa 1-216) was cosedimentated with F-actin, whereas others were not (Fig.  2B). Therefore, the amino terminus of IQGAP1 (aa 1-216) containing the CHD is sufficient for the F-actin binding. To examine whether Cdc42 affects this binding, we used Myc-IQGAP1 (aa 1-1657) because Cdc42 does not interact with IQGAP1-N-1 (aa 1-863) but interacts with Myc-IQGAP1 (aa 1-1657) (14). The similar assay was performed in the presence or the absence of either GTP␥S⅐GST-Cdc42 or GDP⅐GST-Cdc42. Neither GTP␥S⅐GST-Cdc42 nor GDP⅐GST-Cdc42 affected the F-actin binding activity of Myc-IQGAP1 (aa 1-1657) (data not shown). We also examined whether IQGAP1 interacts with globular actin using a GST-IQGAP1-N-1 (aa 1-863) affinity column chromatography and found that IQGAP1 did not interact with globular actin (data not shown).
Next we examined whether IQGAP1 cross-links the F-actin. Actin-based viscosity was measured by the falling ball viscometry assay (29). Actin-based viscosity was markedly increased in the presence of GST-IQGAP1 (aa 1-1657) and GST-IQGAP1-N-1 (aa 1-863) in a dose-dependent manner compared with 2 Identification of p41 will be described elsewhere (study in progress).

FIG. 1. Affinity purification of IQGAP1-interacting molecules.
A, the bovine brain cytosol was loaded onto a glutathione-Sepharose column coated with the indicated GST fusion proteins. The bound proteins were eluted by the addition of Buffer A containing 500 mM NaCl. Aliquots of the eluates were resolved by SDS-PAGE followed by silver staining. The arrow and arrowhead denote the positions of p43 and p41, respectively. B, aliquots of the eluates were resolved by SDS-PAGE followed by Western blotting with anti-actin antibody. The arrow denotes the position of actin. The results shown are representative of three independent experiments. that in the absence of IQGAP1 (Fig. 3A), suggesting that IQGAP1 cross-links F-actin. The shorter fragments, GST-IQGAP1-N-2 (aa 1-216), GST-IQGAP1-N-3 (aa 216 -683), and GST-IQGAP1-M-1 (aa 521-914), did not increase the viscosity.
Actin-cross-linking proteins, such as filamin (19) and ␣-actinin (33), form oligomers, and the oligomerization enables them to cross-link F-actin (25). Therefore, we examined whether IQGAP1 can form oligomers by gel filtration analysis. Myc-IQGAP1 appeared as broad major and minor peaks, corresponding to the molecular masses of about 300 and 500 kDa, respectively (Fig. 5). These peaks may contain monomers, dimers, and trimers, suggesting that IQGAP1 can form oligomers. GTP␥S⅐GST-Cdc42 but not GDP⅐GST-Cdc42 markedly shifted the peak of IQGAP1, corresponding to the molecular mass of about 600 kDa. GTP␥S⅐GST-Cdc42 alone was eluted as a dimer. Judging from the molecular mass of a dimer of GTP␥S⅐ GST-Cdc42 (about 100 kDa), it is likely that GTP␥S⅐GST-Cdc42 enhances the oligomerization of IQGAP1. These results strongly  4. Oligomerization of IQGAP1 through IQGAP1-N-3 (aa  216 -683). A, the bovine brain cytosol was loaded onto a glutathione-Sepharose column coated with indicated GST fusion proteins. The bound proteins were eluted by the addition of Buffer A containing 500 mM NaCl. Aliquots of the eluates were resolved by SDS-PAGE followed by Western blotting with anti-IQGAP1 antibody. The arrow denotes the position of IQGAP1. B, effects of guanine nucleotides (10 M) or Cdc42 (3 M each) on interaction of GST-IQGAP1-N-3 (aa 216 -683) with full-length IQGAP1 was examined. The bovine brain cytosol was mixed at 4°C for 1 h with or without GDP, GTP␥S, GDP⅐Cdc42, or GTP␥S⅐Cdc42 and loaded onto a glutathione-Sepharose column coated with GST-IQGAP1-N-3 (aa 216 -683). Aliquots of the eluates were resolved by SDS-PAGE followed by Western blotting with anti-IQGAP1 antibody. The arrow denotes the positions of IQGAP1. The results shown are representative of three independent experiments. suggest that the oligomerization of IQGAP1 leads to crosslinking the actin filament. DISCUSSION We have previously found that IQGAP1 accumulated at the insulin-and Rac1 val12 -induced membrane ruffling area in KB cells, where cortical actin filament was observed (14). In this study, we found that IQGAP1 directly interacts with F-actin and cross-links the actin filament. Cdc42 and Rac1 have been shown to regulate the filopodia and lamellipodia formation through the reorganization of actin filament meshwork (1,2). Cross-linking of F-actin may be critical for the filopodia and lamellipodia formation. Our result strongly suggests that IQGAP1 plays an important role in these processes through the regulation of actin filament.
The target molecules for Cdc42 and Rac1 have thus far been identified to be PAK (6 -8), WASP (9,10), and IQGAP1 (11)(12)(13)(14). Recently, it is reported that microinjection of activated PAK induces the filopodia formation and membrane ruffling in Swiss 3T3 cells (34). Another group reported that expression of constitutively activated PAK induces dynamic morphological changes (35). However, these results do not exclude the possibility that other targets are involved in the filopodia and lamellipodia formation. WASP (9, 10) and its isoform, N-WASP (16), are shown to alter the actin filament localization and to sever the F-actin, respectively. On the basis of these observations, it is plausible that IQGAP1, together with PAK and WASP, plays an important role in the reorganization of actin filament.
Actin-cross-linking proteins, such as ␣-actinin (20), filamin (22), and fimbrin (23) are believed to bridge F-actin and consequently cross-link the F-actin (25). Although IQGAP1 has a CHD in its amino terminus (aa 1-216) and this domain is sufficient for the interaction with F-actin, this domain lacked the F-actin-cross-linking activity. The IQGAP1-N-1 (aa 1-863) had the F-actin binding and cross-linking activities. Additionally, oligomerization of IQGAP1 is mediated by its aminoterminal portion (aa 216 -683). On the basis of these observations, it is likely that the CHD is sufficient for the binding to F-actin and the portion (aa 216 -683) is responsible for the oligomerization of IQGAP1. Thus, IQGAP1 can cross-link the actin filament in a similar fashion to other actin-cross-linking proteins.
GTP␥S⅐GST-Cdc42 enhanced this IQGAP1 activity when 300 nM of IQGAP1 was used, although Cdc42 did not affect the activity of IQGAP1 when more than 750 nM of IQGAP1 was used. This result suggests that Cdc42 caused a leftward shift of the curve of the IQGAP1 activity. The concentration of endogenous IQGAP1 was calculated to be about 300 nM in cultured cells, such as MTD-1A epithelial cells (data not shown). At this concentration of IQGAP1, Cdc42 can further enhance the crosslinking activity of IQGAP1. Therefore, it is plausible that IQGAP1 physiologically cross-link the actin filament in a Cdc42dependent manner in vivo.
GTP␥S⅐Cdc42 did not affect the interaction of IQGAP1 with F-actin, whereas GTP␥S⅐Cdc42 enhanced the oligomerization of IQGAP1. This may account for how Cdc42 induces the crosslink of the actin filament through IQGAP1. The mechanism by which GTP␥S⅐Cdc42 elicits the oligomerization of IQGAP1 is not known at present. A possible explanation is that GTP␥S⅐Cdc42 affects the conformation of IQGAP1 and subsequently promotes the oligomerization of IQGAP1 through the amino-terminal portion (aa 216 -683) of IQGAP1. However, this possibility may be less likely because neither GTP␥S nor GTP␥S⅐Cdc42 affected the interaction of bovine IQGAP1 with the amino-terminal portion (aa 216 -683) of IQGAP1. Alternatively, Cdc42 itself may make oligomers so that IQGAP1-Cdc42 complex consequently forms additional oligomers. It has been shown that Ras and RhoA, members of small GTPases, form oligomers (36,37). If Cdc42 forms oligomers as in the case of Ras and RhoA, this possibility is likely. However, a further study to address this issue is necessary.
Very recently, during the revision of this manuscript, IQGAP1 has been shown to bind F-actin and cross-link F-actin (38). Our result is consistent with this observation. We showed here that GTP␥S⅐GST-Cdc42 enhanced the F-actin-cross-linking activity of IQGAP1, possibly through enhancing the oligomerization of the IQGAP1-Cdc42 complex. Therefore, IQGAP1 appears to be a key molecule for the actin cytoskeletal reorganization regulated by Cdc42. FIG. 5. Gel filtration chromatography of IQGAP1. Purified Myc-IQGAP1 (25 g) by calmodulin-Sepharose 4B column chromatography was incubated at 4°C for 1 h with or without either GDP⅐GST-Cdc42 or GTP␥S⅐GST-Cdc42 (100 g each). The mixture was loaded onto a Superose 6 HR 10/30 gel filtration column, and then 0.25-ml fractions of each were collected. These fractions were pooled and analyzed by Western blotting. The relative amounts of IQGAP1 were determined by densitometry. f, with GDP⅐GST-Cdc42; E, with GTP␥S⅐GST-Cdc42; q, without GST-Cdc42. The results shown are representative of three independent experiments.